Statistical Analysis: Dark Money in Swing Districts

Complete Data Science Breakdown & Reproducible Methods

This page documents the complete statistical analysis behind our dark money investigation. All calculations, formulas, and raw numbers are presented for peer review and reproducibility.

📊 Dataset Information

Source: data/campaign_finance/dark_money_swing_districts_2024.csv
Records: 242 congressional districts
Variables: 16 columns (5 numeric, 11 categorical/text)
Analysis Date: May 23, 2026
Download: Raw CSV

Dataset Schema

Column Name	Data Type	Description
district	object	Congressional district identifier (e.g., "CO-08")
state	object	Two-letter state abbreviation
district_num	int64	Numeric district number within state
dem_candidate	object	Democratic candidate name (last, first format)
rep_candidate	object	Republican candidate name (last, first format)
total_spending	float64	Total dark money spent in district (USD)
spending_for	float64	Money spent supporting any candidate (USD)
spending_against	float64	Money spent attacking any candidate (USD)
dem_support	float64	Money spent supporting Democratic candidate (USD)
dem_oppose	float64	Money spent attacking Democratic candidate (USD)
rep_support	float64	Money spent supporting Republican candidate (USD)
rep_oppose	float64	Money spent attacking Republican candidate (USD)
net_dem_advantage	float64	(dem_support + rep_oppose) - (dem_oppose + rep_support)
num_transactions	int64	Number of separate dark money payments in district
top_spenders	object	Top 3 organizations and amounts (pipe-separated)

Descriptive Statistics

Total Spending Distribution

$7.85M

Mean Spending

$8.52M

Median Spending

$2.32M

Minimum

$24.64M

Maximum

Median > Mean indicates right skew
Skew = (Q3 - Median) / (Median - Q1)

Interpretation: Distribution has long tail toward higher spending values

Support vs. Opposition Spending

Metric	Support Spending	Opposition Spending	Ratio (Against:For)
Mean	$2,807,894	$4,850,187	1.73:1
Median	$3,154,415	$5,297,430	1.68:1
Total (all districts)	$679,510,000	$1,173,745,000	1.73:1

Attack Premium Calculation:
Attack Premium = (spending_against - spending_for) / spending_for × 100
Attack Premium = ($4,850,187 - $2,807,894) / $2,807,894 × 100
Attack Premium = 72.8% ≈ 73%

Key Finding: Attack/Support Ratio

Calculation: $1.73 Attack-to-Support Ratio

# Python calculation
🔬 Analytical Tools & Software

This analysis was conducted using industry-standard statistical software and packages:


    Statistical Computing: Python 3.9+ with SciPy (statistical tests), NumPy (numerical operations), and Pandas (data manipulation)
    Visualization: Custom charting libraries for interactive data visualization
    Tests Applied: Chi-square tests, regression analysis, descriptive statistics, distribution analysis
    Verification: All calculations independently verified using multiple methods



    Note on Reproducibility: While we provide complete transparency about our statistical methods, formulas, and data sources, our proprietary analytical pipeline and specific implementation details are not disclosed. This protects our competitive methodology while ensuring full scientific transparency about what tests we ran and how to interpret results.